Marcos Malzone and Elliot Roberts of Solera/Audatex, are at the forefront of the development of photo-based damage assessment applications analyzed by artificial intelligence.
At the Canadian Collision Industry Forum (CCIF) virtual meeting, these two international experts, responsible for product development on all continents, explained what a revolution in collision damage assessment artificial intelligence will bring. This technology is on our doorstep.
Solera, which markets the Audatex estimating platform, is pushing the research so that a photo of the damage on a car is automatically translated into an estimate of repair costs. The car is identified by its VIN number, which specifies the model, year and version. A generated layer of the vehicle allows the software, based on artificial intelligence, to trace the affected areas. The user then only has to remove previous damage and the damage estimate will automatically adjust.
“It’s not the perfect solution, says Roberts, it takes an expert to validate the damage, but it’s an incredible tool that provides a very good basis for estimating the work to be done and the price.” The mileage information will also be used to determine the value of the vehicle prior to the collision, which will be used to determine the next steps, such as a total loss claim.
A Huge Database
To build its estimate, the system, which is currently being validated with European insurers and body shops, is working on a database of one billion images and 350 million estimates from workshops. All of this is linked to the Audatex system, which includes all repair processes according to manufacturers’ standards and is constantly updated. The system was built to take into account the cost of work and parts according to different world markets and adjust the estimates accordingly.
“For insurers, this tool promises great transparency and consistency in results, explains Marcos Malzone. The pilot projects currently underway show a 50% reduction in surcharges in the files analyzed. We are working to validate this automatic process based on our data and archived decisions.” Both parties know that in addition to validating the technology, it will also have to be made acceptable to insurers, suppliers and repairers. For the consumer, who could provide the photos to his insurer himself, the entire claims process could be accelerated. If the estimate is accurate and detailed, the body shop would have the opportunity to order the necessary parts and reserve his team before the car even enters the shop.
According to the developers, the platform, which is being validated with the body shop ecosystem, needs to achieve consensus among the consumer, the insurer, parts and equipment suppliers, and body shops. A comparative study is also underway to see what disparities will be detected between an estimate made by software based on artificial intelligence and one compiled by a human.
“The technological challenge no longer stands in the way. The use of photos that can be interpreted in 20 seconds will change the body shop landscape, Roberts predicts. You can do the analysis remotely, quickly, following a protocol that is satisfactory to all parties.”